Alibaba open-weighted this checkpoint on August 12, 2026, 9 days after Qwen3.8-Max's API debut, extending context to roughly 1 million tokens with FP8, GGUF and NVFP4 builds available the same day. It replaces Qwen3.7-Max as Alibaba's flagship reasoning and agentic-coding model for teams that need self-hosting or multi-cloud deployment instead of a single vendor API.
Qwen3.8-2.4T-A95B is Alibaba Cloud's mixture-of-experts flagship, activating 95 billion of its 2.4 trillion parameters per token and scoring 92.6% on GPQA Diamond. It is the first Qwen-Max-class model Alibaba has released as open weights, aimed at agentic coding and long-horizon reasoning workloads that previously required a closed API.
Provider: Alibaba Cloud · Family: Qwen3.8
More about Alibaba Cloud on HokAI
Context window: 1,010,000 tokens · Max output: 131,072
Input modalities: text, tool-calls · Output: text, tool-calls
About Qwen3.8-2.4T-A95B
Qwen3.8-2.4T-A95B is Alibaba Cloud's largest model release to date, open-weighted on Hugging Face on August 12, 2026. It is the checkpoint behind the QwenCloud-hosted Qwen3.8-Max product, GA'd on August 3, 2026. The architecture is a fine-grained mixture-of-experts transformer: 2.4 trillion total parameters, 95 billion active per token, across 512 experts (10 routed plus 1 shared per token) and 92 layers interleaving linear and full attention in a 3:1 pattern. It succeeds Qwen3.7-Max, announced at the Apsara Summit on May 20, 2026, and is the first Qwen-Max-class flagship Alibaba has open-weighted rather than kept API-only. Alibaba reports 92.6% on GPQA Diamond, 67.7% on SWE-bench Pro, 86.6% on Terminal-Bench 2.1, 86.1% on OSWorld-Verified, 93.0% on PaperBench and 82.8% on IFBench. On the independently tracked DeepSWE 1.1 coding benchmark it scores 56.6, well behind the current top three frontier models: its GPQA-level reasoning strength doesn't fully carry over to agentic coding. AIME 2025, MMLU-Pro and ARC-AGI 2 remain unpublished, and independent verification is limited given the model's age. Native context is 262,144 tokens, extensible to roughly 1,010,000, which is where the "1 million token context" figure in most coverage comes from. Recommended max output is 131,072 tokens for a final answer, though up to 262,144 of that budget can go to visible reasoning traces first, since thinking mode is mandatory and cannot be disabled. Despite the "Max" branding, Qwen3.8-2.4T-A95B is text-only: its Hugging Face model card confirms multimodal input is not supported. It accepts text and tool-call input and returns text and tool-calls, with native OpenAI-compatible function calling, structured output and parallel tool calls. This mirrors Qwen3.7-Max, whose text-only "Max" flagship was often confused with the multimodal "Plus" variant; several writeups already describe Qwen3.8-Max as accepting image and video input, which this checkpoint does not. It ships configurable reasoning: low, high and xhigh effort levels per request. QwenCloud's International (Singapore) endpoint undercuts Qwen3.7-Max's prior per-1M-token rate despite the larger size, with a steep discount on cached input (exact figures in the pricing FAQ below). The Mainland China endpoint runs an estimated 60-70% cheaper, and OpenRouter's routed price runs lower still. A 1M-in/200K-out coding-agent run costs about $3.20; a 1,000-turn support chat averaging 2K in/500 out per turn runs about $7.00. Open weights ship on Hugging Face as safetensors, FP8, GGUF (Unsloth) and NVFP4 (RadixArk) builds. Day-0 serving came from SGLang, vLLM, NVIDIA NIM/Dynamo and Fireworks AI, alongside the official QwenCloud API and Alibaba Model Studio. NVIDIA reports over 4,000 tokens/sec per GPU and 350 tokens/sec per user on its GB300 NVL72 reference system, a figure specific to that deployment rather than any hosted API's guaranteed rate. No AWS Bedrock, Google Vertex or Azure availability is confirmed yet. Alibaba has not published a safety system card, training data cutoff, or named red-teaming partners for this model. Its custom license, not Apache 2.0, requires products over 100 million monthly active users to display the model name in their UI, and "Model as a Service" or "AI Work Assistant" businesses over $50 million in trailing 12-month revenue to sign a separate commercial license before commercial use. It suits teams building self-hosted or multi-cloud agentic coding and long-document pipelines who want frontier-scale reasoning without single-vendor lock-in, and who have the multi-GPU infrastructure a 2.4T-parameter checkpoint needs even at FP8 or NVFP4. Teams needing native image or video understanding should use Qwen's separate VL line instead; teams needing a published system card for compliance may prefer GPT-5.6 or Claude Opus 4.8, which document red-teaming partners this model lacks.
Pricing
QwenCloud's International (Singapore) endpoint prices Qwen3.8-2.4T-A95B at $2.00 per 1M input tokens and $6.00 per 1M output tokens, with cached input at $0.25 per 1M (an 8x discount). The Mainland China (Beijing) endpoint runs an estimated 60-70% cheaper for the same model. OpenRouter's routed pricing has floated slightly lower, around $1.80/$5.40 per 1M, depending on which upstream host serves the request. No separate batch-API discount has been published for this model.
Key Features
- Configurable Reasoning Effort: Switches between low, high and xhigh reasoning modes per request, trading inference cost for depth on hard reasoning and coding tasks.
- 1M-Class Context Window: Native 262,144-token context extends to roughly 1,010,000 tokens, enough for large codebases or multi-session agent transcripts in one call.
- Open Weights in Four Formats: Released on Hugging Face as full safetensors, FP8, GGUF (via Unsloth) and NVFP4 (via RadixArk) checkpoints for self-hosted deployment.
- Fine-Grained MoE Routing: 512 total experts with 10 routed plus 1 shared expert active per token, keeping inference cost tied to the 95B active count rather than the full 2.4T parameters.
- Day-0 Inference Stack Support: Four independent serving stacks added same-day support rather than waiting on community ports, with NVIDIA's reference hardware reporting over 4,000 tokens/sec per GPU.
Pros
- First Qwen-Max-class model Alibaba has open-weighted, at $2.00/$6.00 per 1M tokens on the official API, undercutting its own Qwen3.7-Max predecessor's $2.50/$7.50 rate.
- Scores 92.6% on GPQA Diamond, ahead of the prior Qwen3.7-Max flagship on graduate-level reasoning.
- Ships same-day in FP8, GGUF and NVFP4 quantized formats plus Fireworks/SGLang/vLLM/NVIDIA NIM serving support, so self-hosting doesn't require waiting on the community to catch up.
Cons
- DeepSWE coding score of 56.6 trails GPT-5.6 Sol (73), Gemini 5 (70) and Claude Opus 4.8 (59), so the GPQA lead doesn't carry through to every coding benchmark.
- Text-only despite the 'Max' branding: no image, audio or video input, unlike what several third-party writeups imply.
- No published training data cutoff date or safety system card as of August 2026, which complicates enterprise compliance review.
- Custom license requires a separate commercial agreement above $50M in trailing 12-month revenue for hosting or AI coding/office-assistant products, unlike a plain Apache 2.0 release.
Benchmarks
- ifbench: 82.8
- paperbench: 93
- deepswe 1 1: 56.6
- gpqa diamond: 92.6
- swe bench pro: 67.7
- osworld verified: 86.1
- terminal bench 2 1: 86.6
Frequently Asked Questions
How much does Qwen3.8-2.4T-A95B cost per 1M tokens?
Through QwenCloud's international (Singapore) endpoint, Qwen3.8-2.4T-A95B costs $2.00 per 1M input tokens and $6.00 per 1M output tokens, with cached input billed at $0.25 per 1M tokens, an 8x discount. Alibaba's Mainland China endpoint runs an estimated 60-70% cheaper for the same model. That undercuts Qwen3.7-Max's prior $2.50/$7.50 rate, though OpenRouter's routed pricing has floated slightly lower, around $1.80/$5.40, depending on the upstream host.
How does Qwen3.8-2.4T-A95B compare on benchmarks vs GPT-5.6 and Gemini 5?
Alibaba reports 92.6% on GPQA Diamond and 67.7% on SWE-bench Pro for Qwen3.8-2.4T-A95B, ahead of its own Qwen3.7-Max predecessor on graduate-level reasoning. On the independent DeepSWE coding benchmark it scores 56.6, trailing GPT-5.6 Sol (73), Gemini 5 (70), and Claude Opus 4.8 (59), so its coding edge is narrower than the GPQA number suggests. AIME 2025 and MMLU-Pro scores were not published as of this writing.
Is Qwen3.8-2.4T-A95B open source or proprietary?
Qwen3.8-2.4T-A95B ships as open weights on Hugging Face in safetensors, FP8, GGUF and NVFP4 formats, the first Qwen-Max-class checkpoint Alibaba has open-weighted. It carries a custom Qwen license, not Apache 2.0: products with over 100 million monthly active users must display the model name in their UI, and businesses earning over $50 million in trailing 12-month revenue from model-hosting or AI coding/office-assistant products need a separate commercial license.
Does Qwen3.8-2.4T-A95B train on user data?
Alibaba has not published a training data cutoff date or a system card for Qwen3.8-2.4T-A95B as of August 2026. For hosted use through QwenCloud, Alibaba's general terms of service govern data handling; no zero-retention or enterprise data-processing agreement specific to this model has been publicly documented yet.
Who is Qwen3.8-2.4T-A95B best for and who should avoid it?
Qwen3.8-2.4T-A95B suits teams running self-hosted or multi-cloud agentic coding pipelines who want frontier-scale reasoning without single-vendor API lock-in. Teams needing native image or video input should look elsewhere: the checkpoint is text-only despite the 'Max' branding. Teams needing a published safety system card for compliance review may prefer Claude Opus 4.8 or GPT-5.6, which both document red-teaming partners.